Xenial build/deployment failure to initialise deployment scripts ("Error installing dpl")

[jrg]

Our builds have started failing, sometimes, with a ruby gem related error. I believe it’s when the “deploy” section prepares to be run as it’s not from our own scripts. The build log records:

$ rvm use 2 --fuzzy do ruby -S gem install dpl -v ''
ERROR:  Error installing dpl:
	The last version of dpl (>= 0) to support your Ruby & RubyGems was 1.10.16. Try installing it with `gem install dpl -v 1.10.16`
	dpl requires Ruby version >= 3.1. The current ruby version is 2.7.6.219.

The “Build system information” is identified as:

Build language: java
Build dist: xenial
...
travis-build version: 7c7c1d60

The versions of every tool in the Build System information section are identical, including the alleged default ruby version being ruby 3.3.5 (2024-09-03 revision ef084cc8f4) [x86_64-linux].

(edited to reflect realisation that Build ID is not the ID of the build distribution image)

[jrg]

ok, I was initially assuming that the Build ID was the build of the Xenial image, but I’ve now realised that it’s the actual build id of our builds.

The “travis-build version” is the same both when it works and when it doesn’t:

travis-build version: 7c7c1d60

So, when it works, the dpl command in the log looks like:

$ rvm use $(travis_internal_ruby) --fuzzy do ruby -S gem install dpl -v 2.0.5.1

[acacopardo]

We have the same issue.

[kpatidarsecben]

We are facing the same issue. We have dist as jammy. It is failing on deploy section.

[ArtemPWG]

Getting the same issue with xenial dist, jobs failing randomly on deploy section

[lily-vu-goldenowl]

Hi, I got the same issue today.
Currently, the travis-ci runs this command as the default.

rvm use 2 --fuzzy do ruby -S gem install dpl -v ''

It used to successfully deploy with another default command

rvm use $(travis_internal_ruby) --fuzzy do ruby -S gem install dpl -v 2.0.5.1